Real time data collection and analysis would substantially strengthen public health surveillance at both regional and national levels. Although symptom and diagnostic data are stored in administrative health care databases, there currently are no automated systems for integrating this data so that abnormal patterns of disease can be detected in a timely manner, a problem highlighted by recent bioterrorist attacks. We aim to address this by working with an existing and rapidly expanding data acquisition infrastructure to develop the analytic tools needed to recognize disease clusters as soon as possible once patients begin appearing at health care sites. We have already established a surveillance network by virtually integrating multiple hospital emergency department (ED) databases in real time. This provides a picture of regional population patterns of disease. Furthermore, at one of these hospitals, we have assembled retrospective databases with many years of historical data critical for establishing normality for the analytic models. Thus, the goal of this project is to strengthen surveillance systems by using readily available information in realtime, thereby increasing the power of these systems to detect disease outbreaks sooner. Because of the urgent need to quickly identify victims of biowarfare agents such as anthrax, we will concentrate on identifying clusters of patients presenting with respiratory syndromes. The proposed study will first establish normal patterns of disease and then build models that enable the detection of deviations from these patterns with a minimum of false alarms. Once these methods are established, they can be applied to any other meaningful set of syndromes.